Multiply robust matching estimators of average and quantile treatment effects

TL;DR

This paper introduces double score matching (DSM), a new method in causal inference that combines propensity and prognostic scores with multiple models to ensure robustness against model misspecification.

Contribution

It proposes a novel double score matching estimator that achieves multiple robustness by incorporating multiple candidate models for propensity and prognostic scores.

Findings

DSM estimator is consistent if any model for propensity or prognostic score is correct.

The method enhances causal effect estimation robustness.

The approach extends traditional propensity score matching with multiple model considerations.

Abstract

Propensity score matching has been a long-standing tradition for handling confounding in causal inference, however requiring stringent model assumptions. In this article, we propose double score matching(DSM) for general causal estimands utilizing two balancing scores including the propensity score and prognostic score. To gain the protection of possible model misspecification, we posit multiple candidate models for each score. We show that the de-biasing DSM estimator achieves the multiple robustness property in that it is consistent for the true causal estimand if any model of the propensity score or prognostic score is correct.